NVIDIA Research (Efficient AI Team & Singapore Lab) · Text-to-Video

SANA-Video 2.0

NVIDIA hybrid video diffusion transformer at 5B and 14B scales that replaces most quadratic attention with gated linear attention, keeping 25% softmax anchors, and trains from scratch rather than linearizing a pretrained model. Scores VBench 84.30 and runs 720p/5s in 13.06s on one H100. Checkpoints not yet released; samples marked coming soon.